Email: Password: Remember Me | Create Account (Free)

Back to Subject List

Old thread has been locked -- no new posts accepted in this thread
???
11/12/05 14:31
Read: times


 
#103621 - Filter providing two time constants
Responding to: ???'s previous message
Thakkar said:
I am designing a weight scale using ic7135.The samples very between three to four, or sometimes five values. What would be the best method to avarage out the samples, so that result seems to be steady?

There can be some reasons for noisy readings: Noisy reference voltage, poor ground routing intoducing hum loops, injected hum due to a mains transformer in close distance to ICL7135, signal noise at input, etc.
Mostly the noise comes from input signal noise, which can be rejected to some degree by a suited low pass filter. Standardly you should use a low pass filter consisting of 100kOhm and 100nF, as it's suggested in datasheet. But if this does not present enough rejection, then corner frequency should be lowered. Unfortunately, settling time increases then, means it takes longer to get a stable reading and for big changes of input signal settling time can become unacceptably high.
The following circuit offers both, fast settling time for big changes of input signal and at the same time improved noise filtering due to lowered corner frequency:


The circuit works as follows: If a big change of input signal arrives, then one of the Schottky diodes becomes conducting whenever the difference between input and output voltage of the filter is higher than about 0.2V. In this case only the 1kOhm resistor forms the time constant of low pass filter. So, the capacitor is charged up much faster, than with the 100kOhm resistor alone.
If the difference between input and output voltage of filter falls under 0.2V, the Schottky stops conducting and now the 100kOhm resistor forms the much bigger time constant of low pass filter, resulting in improved noise filtering.

Take care, only polypropylene capacitors should be used for the caps at ICL7135!!

Kai

List of 16 messages in thread
TopicAuthorDate
Weight Scale Avarging method            01/01/70 00:00      
   Filter providing two time constants            01/01/70 00:00      
      median filter, simulate algo on PC            01/01/70 00:00      
         Periodic interference can make them fail            01/01/70 00:00      
            Acknowlegded.            01/01/70 00:00      
               Agreed            01/01/70 00:00      
                  50Hz sampling            01/01/70 00:00      
                     Anti-aliasing filter will help            01/01/70 00:00      
                        what this is all about            01/01/70 00:00      
   Moving average window.            01/01/70 00:00      
      Average on condition            01/01/70 00:00      
         is that not SOP (standard operating proc            01/01/70 00:00      
            Hallelujah!            01/01/70 00:00      
   meaningless            01/01/70 00:00      
   What are you using that one for in a wei            01/01/70 00:00      
   hello from chhatraliamd for weight scale            01/01/70 00:00      

Back to Subject List